Introduction
In the context of contemporary climate change, rising sea levels and human pressures compromise the resilience of deltaic environments. A deeper understanding of the processes underlying delta formation is crucial for their preservation and sustainable development.
Deltaic morphology arises from the interplay between natural and human influences, mediated by the fluvial system’s sediment supply, discharge, and intrinsic dynamics (Galloway, Reference Galloway1975; Boyd et al., Reference Boyd, Dalrymple and Zaitlin1992; Bhattacharya and Giosan, Reference Bhattacharya and Giosan2003; Bravard et al., Reference Bravard, Lippmann-Provansal, Arnaud-Fassetta, Chabbert, Gaydou, Dufour, Richard, Valleteau, Melun and Passy2008; Zolezzi et al., Reference Zolezzi, Luchi and Tubino2012). Variations in detrital input and hydrological parameters promote progradation or erosion (Passega, Reference Passega1957; Wright, Reference Wright1977; Schumm, Reference Schumm, Ethridge and Flores1981, Reference Schumm1985; Boyd et al., Reference Boyd, Dalrymple and Zaitlin1992; Dalrymple et al., Reference Dalrymple, Zaitlin and Boyd1992; Bhattacharya and Giosan, Reference Bhattacharya and Giosan2003). These processes are modulated by climatic (Bravard, Reference Bravard1989; Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Goodbred, Reference Goodbred2003; Clift and Jonell, Reference Clift and Jonell2021), eustatic (Stanley and Warne, Reference Stanley and Warne1994; Vella et al., Reference Vella, Fleury, Raccasi, Provansal, Sabatier and Bourcier2005; Fanget et al., Reference Fanget, Berné, Jouet, Bassetti, Dennielou, Maillet and Tondut2014), and tectonic variations (Colella, Reference Colella and Nemec1988). Human activities, such as deforestation and hydraulic engineering, have further altered sediment budgets and deltaic dynamics (Keesstra et al., Reference Keesstra, Van Huissteden, Vandenberghe, Van Dam, De Gier and Pleizier2005; Ericson et al., Reference Ericson, Vörösmarty, Dingman, Ward and Meybeck2006; Syvitski and Saito, Reference Syvitski and Saito2007; Bravard et al., Reference Bravard, Goichot and Tronchère2014; Anthony et al., Reference Anthony, Brunier, Besset, Goichot, Dussouillez and Nguyen2015).
The Rhône Delta (Camargue, France) offers a pertinent case study, having undergone successive avulsions and rapid morphological changes during the late medieval and early modern periods (Arnaud-Fassetta, Reference Arnaud-Fassetta1998; Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Rey et al., Reference Rey, Lefèvre and Vella2005; Vella et al., Reference Vella, Fleury, Raccasi, Provansal, Sabatier and Bourcier2005). Although these evolutionary phases are partly documented by historical sources (Pichard, Reference Pichard1995; Pichard et al., Reference Pichard, Provansal and Sabatier2014; authors of this study, unpublished data), significant gaps remain regarding its modern period (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Arnaud-Fassetta, Reference Arnaud-Fassetta2003), which this study addresses.
The Little Ice Age (LIA; fourteenth to mid-nineteenth century CE) was a period of climatic cooling, driven by reduced solar activity (Bond et al., Reference Bond, Kromer, Beer, Muscheler, Evans, Showers, Hoffmann, Lotti-Bond, Hajdas and Bonani2001; Owens et al., Reference Owens, Lockwood, Hawkins, Usoskin, Jones, Barnard and Scott2017), increased volcanic activity towards the end of the LIA (Miller et al., Reference Miller, Geirsdóttir, Zhong, Larsen, Otto‐Bliesner, Holland and Bailey2012; Brönnimann et al., Reference Brönnimann, Franke, Nussbaumer, Zumbühl, Steiner, Trachsel and Hegerl2019), and changes in thermohaline circulation (Broecker, Reference Broecker2000). This resulted in more frequent and higher-magnitude flood events across European river systems (Carozza et al., Reference Carozza, Puig, Odiot, Valette and Passarrius2012; Schulte et al., Reference Schulte, Peña, Carvalho, Schmidt, Julià, Llorca and Veit2015; Perșoiu and Perșoiu, Reference Perșoiu and Perșoiu2019). Within this context, the Bras de Fer distributary remained active from 1587 to 1711 CE, undergoing significant floods (Rossiaud, Reference Rossiaud1994; Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999) and episodes of intense lateral accretion and progradation that contributed to one of the delta’s latest expansion phases (authors of this study, unpublished data).
This study aims to reconstruct the morphodynamic evolution of the Grande Ponche meander—the most spatially extensive meander of the Bras de Fer— during the LIA through an integrated geomorphological, sedimentological, and geochemical approach. By examining extensive crevasse splay deposits and meander migration patterns, it seeks to highlight how short-term climatic variability affected deltaic morphology.
This study first outlines the palaeoenvironmental and hydrogeomorphological framework of the Bras de Fer channel, together with the geological and geomorphological context of the Rhône delta. It then examines the principal processes that shaped the channel’s evolution during the LIA.
Palaeoenvironmental and hydrogeomorphological framework of the Bras de Fer channel
Geological setting
During the Holocene, the Rhône delta underwent several phases of progradation (L’Homer, Reference L’Homer1975; Aloisi, Reference Aloïsi1986; Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Rey et al., Reference Rey, Lefèvre and Vella2005, Reference Rey, Lefevre and Vella2009; Vella et al., Reference Vella, Fleury, Gensous, Labaune and Tesson2008) controlled by fluctuations in sea level (Vella et al., Reference Vella, Fleury, Raccasi, Provansal, Sabatier and Bourcier2005; Stanford et al., Reference Stanford, Hemingway, Rohling, Challenor, Medina-Elizalde and Lester2011; Fanget et al., Reference Fanget, Berné, Jouet, Bassetti, Dennielou, Maillet and Tondut2014; Fig. 1). Sediments accumulated over a Pleistocene gravel substrate following the deceleration of sea-level rise in the Mediterranean between 8000 and 6000 cal yr BP (Stanley and Warne, Reference Stanley and Warne1994). The Saint-Férréol and Ulmet lobes, comprising the ancient delta, emerged between approximately 4000 and 2000 cal yr BP (Vella et al., Reference Vella, Fleury, Raccasi, Provansal, Sabatier and Bourcier2005). While the eastern and southern sectors of the delta plain developed during the medieval and modern periods (post-500 CE; Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta and Provansal1993), the southwestern part dates back to late antiquity and medieval times (post-100 BCE; Rey et al., Reference Rey, Lefèvre and Vella2005, Reference Rey, Lefevre and Vella2009).
Holocene evolution of the three main Rhône delta lobes, based on palaeo-shoreline dating (L’Homer et al., Reference L’Homer, Bazile, Thommeret and Thommeret1981; Rey et al., Reference Rey, Lefèvre and Vella2005; Vella et al., Reference Vella, Fleury, Gensous, Labaune and Tesson2008). This study focuses on the eastern channel, the Bras de Fer (highlighted in blue with a thick line). Satellite imagery: ESRI (2023).
Geomorphological setting
Understanding the geomorphological evolution of the Rhône delta is fundamental to interpreting the sedimentary processes that shaped it. Extensive research has explored the geomorphology and sedimentology of the delta plain for the Holocene period (Duboul-Razavet and Duplaix, Reference Duboul-Razavet and Duplaix1956; Suanez et al., Reference Suanez, Bruzzi and Arnoux-Chiavassa1998; Arnaud-Fassetta, Reference Arnaud-Fassetta1998; Provansal et al., Reference Provansal, Arnaud-Fassetta, Vella, Landuré and Pasqualini2004; Arnaud-Fassetta et al., Reference Arnaud-Fassetta, Bruneton, Berger, Beaudouin, Boes and Provansal2005; Vella et al., Reference Vella, Fleury, Raccasi, Provansal, Sabatier and Bourcier2005, Reference Vella, Fleury, Gensous, Labaune and Tesson2008; Martinez et al., Reference Martinez, Deschamps, Amorosi, Jouet, Vella, Ducret and Berger2024). Despite the availability of historical and archival data (Rossiaud, Reference Rossiaud1994; Caritey, Reference Caritey1995; Pichard, Reference Pichard1995; Landuré and Pasqualini, Reference Landuré and Pasqualini2004; Pichard and Roucaute, Reference Pichard and Roucaute2014; Pichard et al., Reference Pichard, Provansal and Sabatier2014), detailed information on the medieval and modern geomorphology remains limited (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Rey et al., Reference Rey, Lefèvre and Vella2005, Reference Rey, Lefevre and Vella2009).
Holocene investigations have primarily targeted Roman antiquity (100 BCE–500 CE; Arnaud-Fassetta, Reference Arnaud-Fassetta2000, Reference Arnaud-Fassetta2002; Leveau, Reference Leveau2014; Vella et al., Reference Vella, Tomatis and Sivan2014, Reference Vella, Landuré, Long, Dussouillez, Fleury, Tomatis, Sivan, Sanchez and Jézégou2016) and the industrial period (post-1800 CE; Caritey, Reference Caritey1995; Sabatier, Reference Sabatier2001; Antonelli, Reference Antonelli2002; Maillet, Reference Maillet2005; Sabatier and Anthony, Reference Sabatier, Anthony, Randazzo, Jackson and Cooper2015; Boudet et al., Reference Boudet, Sabatier and Radakovitch2017), leaving the LIA (∼1300–1850 CE) comparatively understudied. Our research addresses this gap by reconstructing the evolution of the Bras de Fer distributary (1587–1711 CE) during the LIA, a period of active sediment supply and lobe construction (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Vella et al., Reference Vella, Fleury, Raccasi, Provansal, Sabatier and Bourcier2005).
The delta’s meandering style is driven by discharge, sediment load, topography, and from modern times, anthropogenic interventions (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Arnaud-Fassetta, Reference Arnaud-Fassetta2002; Vella et al., Reference Vella, Fleury, Raccasi, Provansal, Sabatier and Bourcier2005, Reference Vella, Tomatis and Sivan2014, Reference Vella, Landuré, Long, Dussouillez, Fleury, Tomatis, Sivan, Sanchez and Jézégou2016; Bravard, Reference Bravard2010). From the Middle Ages onwards, deforestation and changes in land use increased sediment supply (Bravard, Reference Bravard1989, Reference Bravard2010). In the twentieth century CE, however, hydraulic constructions progressively reduced erosion and sediment transport, curbing delta progradation (Maillet et al., Reference Maillet, Vella, Provansal and Sabatier2006; Besset et al., Reference Besset, Anthony and Sabatier2017).
The impact of the LIA on the Rhône delta remains poorly understood, particularly regarding overbank features such as crevasse splays. During this period, the delta exhibited a lobate or elongated morphology, driven by stable sea-level conditions and abundant sediment supply (Vella et al., Reference Vella, Fleury, Raccasi, Provansal, Sabatier and Bourcier2005). Fluvial activity is reflected in overbank deposits, especially crevasse splays formed by levee breaches during floods (Russell, Reference Russell1954; Coleman, Reference Coleman1969; Bridge, Reference Bridge, Posamentier and Walker2006; Yuill et al., Reference Yuill, Khadka, Pereira, Allison and Meselhe2016). Their extent depends on both allogenic (climate, discharge, sediment load) and autogenic factors (slope, sinuosity) (Millard et al., Reference Millard, Hajek and Edmonds2017; Rahman et al., Reference Rahman, Howell and MacDonald2022). The formation of crevasse splays often involves successive periods of flooding (Coleman, Reference Coleman1969; Millard et al., Reference Millard, Hajek and Edmonds2017), promoting progradation and possible avulsion (Smith et al., Reference Smith, Cross, Dufficy and Clough1989; Slingerland and Smith, Reference Slingerland and Smith1998).
In the Rhône delta, crevasse splays were previously documented during Roman times, with fine sandy deposits up to 70 cm thick (Arnaud-Fassetta, Reference Arnaud-Fassetta2002). A contemporary example includes the 2003 CE breach of a dyke in the upstream part of the delta, leading to a crevasse splay covering approximately 429 ± 21 km2 (Arnaud-Fassetta, Reference Arnaud-Fassetta2013).
In the present study, extensive crevasse splays were identified along the outside bank of the Grande Ponche meander (Fig. 2), often exceeding 2 m in thickness. By combining granulometry, geophysics, geochemistry, historical cartography, and remote sensing, we reconstruct the LIA fluvial dynamics of the Rhône Delta at decadal resolution.
(A) Four meanders of the Bras de Fer and associated landscape features, with the study area outlined. Sources: 1, Arnaud-Fassetta (Reference Arnaud-Fassetta1998); 2, Vella et al. (Reference Vella, Landuré, Long, Dussouillez, Fleury, Tomatis, Sivan, Sanchez and Jézégou2016); (B) Pléiades (CNES 2023) satellite image of the study area (A), showing continuous fossil outside overbank deposits of the Grande Ponche meander. Boundaries of lateral deposits (dashed lines) derived from a digital terrain model (DTM). Green dots indicate drill core locations from a previous study conducted by Arnaud-Fassetta (Reference Arnaud-Fassetta1998).
These geomorphological characteristics provide the basis for understanding the fluvial dynamics of the Bras de Fer channel during the LIA. The following section examines the hydrogeomorphological conditions that governed its activity, in particular the role of climatic variability and flood regimes.
Hydrogeomorphological settings of the Bras de Fer channel
During the LIA (∼1300–1850 CE), the Rhône basin underwent significant hydrogeomorphological changes. Wetter phases favoured meander expansion and overbank flooding, while drier periods encouraged sediment deposition within channels, forming bars and islets (Bravard, Reference Bravard1989, Reference Bravard2010; Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; authors of this study, unpublished data). Historical cartography documents these fluctuations in delta geomorphology (Caritey, Reference Caritey1995; Arnaud-Fassetta, Reference Arnaud-Fassetta1998; Pichard et al., Reference Pichard, Provansal and Sabatier2014; authors of this study, unpublished data).
The Bras de Fer was active during a period of increased precipitation and flood frequency across the French Mediterranean region (Pichard, Reference Pichard1995). Enhanced runoff transported larger sediment volumes, fostering morphodynamic instability, channel shifts (Arnaud-Fassetta, Reference Arnaud-Fassetta2003; Carozza et al., Reference Carozza, Puig, Odiot, Valette and Passarrius2012), delta-lobe progradation (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Pichard et al., Reference Pichard, Provansal and Sabatier2014), and morphological transformations (Bravard, Reference Bravard1989, Reference Bravard2010). Similar patterns were observed in other European river systems (Bellotti et al., Reference Bellotti, Caputo, Davoli, Evangelista, Garzanti, Pugliese and Valeri2004; Perșoiu and Perșoiu, Reference Perșoiu and Perșoiu2019; Ruiz-Pérez and Carmona, Reference Ruiz-Pérez and Carmona2019) and beyond (Seltzer and Rodbell, Reference Seltzer and Rodbell2005).
Major flood events triggered the 1587 CE avulsion of the Grand Passon channel, redirecting flow through the Bras de Fer distributary (Rossiaud, Reference Rossiaud1994; Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999). This shift was reinforced by enhanced sedimentation, resulting from Alpine deforestation, overgrazing, and intense flood activity (Sclafert, Reference Sclafert1959; Ladurie, Reference Ladurie1967; Bravard, Reference Bravard1989; Arnaud-Fassetta, Reference Arnaud-Fassetta1998; Pichard et al., Reference Pichard, Provansal and Sabatier2014).
By the late 1600s and early 1700s CE, a high frequency of extreme floods (Pichard, Reference Pichard1995) remobilised sediments, driving delta progradation until the 1711 CE avulsion. This event interrupted sediment supply to the Bras de Fer mouth, initiating coastal erosion and sediment redistribution (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Pichard et al., Reference Pichard, Provansal and Sabatier2014; authors of this study, unpublished data).
During the LIA, Mediterranean deltas such as the Rhône displayed a fluvial-dominated regime, contrasting with today’s wave-dominated systems (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Bellotti et al., Reference Bellotti, Caputo, Davoli, Evangelista, Garzanti, Pugliese and Valeri2004; Ruiz-Pérez and Carmona, Reference Ruiz-Pérez and Carmona2019).
Building on this hydrogeomorphological context, the next section focuses on the specific characteristics of the Grande Ponche meander, the key study area selected to reconstruct the fluvial dynamics of the Bras de Fer channel during the LIA.
Study area
This study examines the morphometric evolution of the Grande Ponche meander and the associated overbank deposits along its outside bank. The area was selected for its well-preserved palaeolandforms of lateral sedimentation (Fig. 2B), which have remained relatively undisturbed due to limited agricultural activity and protection within a regional natural park.
The Grande Ponche is one of four meanders along the Bras de Fer, whose amplitude declines seaward (Fig. 2A). According to Schumm’s (Reference Schumm, Ethridge and Flores1981, Reference Schumm1985) classification, this reach is close to a low-stability meandering channel due to its gentle slope, fine sediment load, wide meander bars, and increased channel width at apices.
The northernmost meander (no. 1, in red in Fig. 2A) is circular and almost symmetrical, situated in an area heavily modified by agriculture and intersected by palaeochannels (Ulmet, Grand Passon, and Bras de Fer).
The Grande Ponche meander (no. 2), which is symmetrical and elongated (Fig. 2A), shows a chute cutoff, downstream shifting, and well-preserved overbank deposits, including topographically pronounced crevasse splays, which likely capture the dynamics of intense flood episodes. These deposits are expected to contain coarser sediments due to higher shear stress and velocities on the outside bank (Clayton and Pitlick, Reference Clayton and Pitlick2007). Sedimentation on the inside bank, documenting the channel’s active phase and infilling, was previously described by Arnaud-Fassetta and Provansal (Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999) (core sample locations in green in Fig. 2B).
These first two meanders likely began forming when the Bras de Fer became the main distributary, or possibly earlier, during the final activity of the Grand Passon. It may have recorded multiple phases of flood intensification and reduction.
The Tampan meander (no. 3) is less developed and slightly asymmetrical, featuring several internal bars (Fig. 2A). Historical cartography suggests it formed later than the Grande Ponche, likely during a short period of large floods in the second half of the seventeenth century CE.
A fourth, subtle meander (no. 4 in Fig. 2A), marks the onset of sinuosity along an otherwise straighter reach, active between the mid-seventeenth and early eighteenth centuries CE, before the Rhône’s 1711 CE avulsion.
Methodology
Morphometric evolution of the Grande Ponche meander using ancient maps and aerial images
Historical maps containing at least four stable landmarks (e.g., towers, farmsteads) were georeferenced using thin plate spline transformations (authors of this study, unpublished data). Early or distorted maps lacking sufficient control points were nonetheless utilised for descriptive purposes. Two undated maps were provisionally dated based on cartographic features and comparison with documented meander dynamics.
Recent spatial data supplemented interpretations, including 1942 aerial photographs (“C2942-0011_1942_NIMES-CAMARGUE” [IGN, 1942]). These photographs were orthorectified into mosaics, capturing morphologies later modified by agriculture.
Morphometric properties and estimation of meander migration
The morphological evolution of the Grande Ponche meander was reconstructed using historical maps (Supplementary Table 1) and 1942 imagery. Standard morphometric parameters were calculated: sinuosity (S), wavelength (λ), amplitude (A), radius of curvature (Rc), and channel width (W) (Leopold and Wolman, Reference Leopold and Wolman1957; Hooke, Reference Hooke1984, Reference Hooke and Shroder2013; Knighton, Reference Knighton1998).
The Rc/W ratio indicates meander dynamics: 4–20 suggests growth, 2–3 denotes active migration, and <2 implies chute cutoff or abandonment (Bagnold, Reference Bagnold1960; Hey, Reference Hey1984; Hooke, Reference Hooke1997). Meander expansion rates were calculated based on changes in Rc across dated intervals, and lateral extension was assessed from apex migration relative to elapsed time. Bankfull discharge (Qb) and annual peak flood discharge (Q1.58) were estimated using Dury’s equations (Reference Dury1955, Reference Dury1976).
Although the fossilised and partially modified state of the channel may introduce uncertainties, these estimates provide valuable insights into its historical dynamics.
Interpretation of overbank deposits using a digital terrain model
A 50-cm-resolution digital terrain model (DTM) was created from a LIDAR HD (IGN 2022) point cloud and processed using R Studio scripts to retain only terrain returns, then interpolated into a triangulated irregular network surface. The original density of 10 points/m2 (compared with 1 point/m2 in previous datasets [SHOM, 2015; IGN, 2018]), allowed detection of subtle features like crevasse splays, levees, and crevasse channels.
The DTM was used to target depositional forms for in situ stratigraphic surveys, sedimentological analyses, and geophysical investigations.
Sedimentary analysis
Stratigraphic data from three trenches, eight cores, and multiple auger samples were used to reconstruct depositional episodes. Among various sedimentary parameters, grain size and isothermal remanent magnetisation (IRM) intensity were selected as key indicators for depositional environments.
Grain-size analysis
Samples were taken at facies changes. Carbonates and organic matter were removed using HCl (35%) and H₂O₂ (30%) (for protocol, see Sharifi et al. [Reference Sharifi, Murphy, Pourmand, Clement, Canuel, Naderi Beni and Lahijani2018] and Lepage et al. [Reference Lepage, Masson, Delanghe and Le Bescond2019]). Grain-size distributions were measured by laser diffraction (Beckman Coulter LS 13 320, Malvern Mastersizer 3000). Both instruments detect particles from ∼0.04 to 2000 μm. The Blott and Pye (Reference Blott and Pye2012) classification scheme was adopted, with clay (<3.91 μm) grouped into a single class due to its low abundance (<10%). Results were presented as bar charts and median grain-size distributions by facies. Depositional hydrodynamics were further characterised using a Passega (Reference Passega1957, Reference Passega1964) CM diagram, which correlates coarsest percentile (C) with median size (M) to characterise depositional hydrodynamics.
Detailed information on the calibration protocols and data variability between instruments can be found in Supplementary Material 1.
IRM intensity
U-channels were magnetised using a 0.6 T Halbach cylinder to impart IRM, then scanned at high resolution (500 μm) with a fluxgate-based device (Demory et al., Reference Demory, Uehara, Quesnel, Rochette, Romey, Tachikawa and Escutia2019). This process mobilises most ferromagnetic particles, mainly iron oxides and hydroxides typical of oxic sediments (Scheidt et al., Reference Scheidt, Egli, Frederichs, Hambach and Rolf2017).
Sensitive to detrital input, IRM serves as a proxy for reconstructing fluvial dynamics. The scanner’s centimetre-scale resolution allows precise correlation of magnetic signals with sedimentary facies, refining environmental reconstructions (Demory et al., Reference Demory, Uehara, Quesnel, Rochette, Romey, Tachikawa and Escutia2019).
Elemental composition and radiography
Elemental composition and sediment radiography were obtained using high-resolution X-ray fluorescence (XRF) scanning (ITRAX, Cox Analytical Systems) on U-channels with a Mo tube as the X-ray source. Scanning at 2 mm intervals (30 kV, 40 mA, 15 s count) provided intensity profiles of terrigenous elements (Fe, Ti, Si, K, Zr) and carbonates (Ca). Radiographic imaging (45 kV, 40 mA, 400 ms exposure, 200 µm resolution) simultaneously delivered density-contrast data, enhancing the detection of sedimentary structures beyond visual descriptions (Croudace et al., Reference Croudace, Rindby, Rothwell and Rothwell2006).
Geochemical proxies supported palaeoenvironmental reconstructions. Zr/Rb ratios served as a grain-size proxy: higher values indicate coarser, flood-transported material (Turner et al., Reference Turner, Jones, Brewer, Macklin and Rassner2015; Talská et al., Reference Talská, Hron and Grygar2021), with zirconium mainly sourced from the Massif Central in the Rhône delta (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999).
Ca/Si distinguished carbonate from siliceous inputs: high Ca values mark carbonate-rich flood deposits, whereas high Si denotes silicate-dominated sources (Schulte et al., Reference Schulte, Peña, Carvalho, Schmidt, Julià, Llorca and Veit2015). Both elements are thus exploited here as tracers of high-energy events.
Si/K acted as a proxy for the siliceous-to-clay content ratio, with higher values indicating sandy sediments and lower ones marking clay-rich deposits. Potassium, associated with clays, signals increased fine-grained sediment supply during floods (Schulte et al., Reference Schulte, Peña, Carvalho, Schmidt, Julià, Llorca and Veit2015).
Subsurface stratigraphy by electrical resistivity tomography and trenches
Electrical resistivity tomography (ERT) is a non-invasive technique for mapping subsurface resistivity variations (Maillet et al., Reference Maillet, Rizzo, Revil and Vella2005; Florsch and Muhlach, Reference Florsch and Muhlach2018), effective in studying alluvial deposits (Maillet et al., Reference Maillet, Rizzo, Revil and Vella2005; Laigre et al., Reference Laigre, Reynard, Arnaud-Fassetta, Baron and Glenz2012; Salomon et al., Reference Salomon, Keay, Strutt, Goiran, Millett and Germoni2016; Bellmunt et al., Reference Bellmunt, Gabas, Macau, Benjumea, Vilà and Figueras2022).
Resistivity is influenced by sediment composition, water content, texture, clay content, and porosity (Rhoades et al., Reference Rhoades, Manteghi, Shouse and Alves1989). In saturated sandy zones, lower values are common, especially near the water table (∼0 m depth).
Nine ERT profiles were acquired across the Grande Ponche overbank deposits using an ABEM Terrameter SAS4000 (Schlumberger-Wenner array, 0.5–1 m electrode spacing), reaching depths of up to 8 m. Data were processed with RES2DINV (Geotomo). Stratigraphic observations from trenches validated and refined the geophysical interpretations.
Radiocarbon dating procedures
Radiocarbon dating was performed on bulk organic matter and gastropod shells from trenches. Calibrations used Calib Rev 8.1.0 software (Reimer et al., Reference Reimer, Austin, Bard, Bayliss, Blackwell, Ramsey and van der Plicht2020) and the IntCal20 curve. Results are expressed in calibrated years BP (2σ range), detailed in Supplementary Table 3.
Results
Morphometric evolution of the Grande-Ponche meander
An analysis of 15 historical maps (Supplementary Table 1) implies that the Grande Ponche meander developed during the Bras de Fer’s period of activity as the main distributary, potentially becoming significant before the avulsion event. The 1593 CE map, the earliest to show the meander, indicates an established channel, possibly initially a secondary one.
Throughout the seventeenth century CE, the meander migrated laterally, evidenced by shifting toponyms and islet formation along the inside bank. After 1688 CE, migration accelerated, with prominent scroll bar development and increasing distance between the active bank and the Grande Ponche farmhouse (Fig. 3, Supplementary Fig. 1).
Evolution of the Grande Ponche meander (1635–1734 CE) based on georeferenced historical maps over 1942 aerial photographs (IGN Remonter le temps, 1942). (A–C) Gradual shift and increasing curvature of the meander (1635–1706 CE). (D) Post-avulsion phase and onset of fossilisation. Core samples from Arnaud-Fassetta (Reference Arnaud-Fassetta1998) are marked in pink.
Georeferencing enabled the repositioning of the former channel on 1942 imagery (Fig. 3), allowing morphometric tracking (Supplementary Table 2). Sinuosity (S) increased from 1.7 in 1635 CE to 2.5 before the chute cutoff; amplitude (A) expanded from 2.8 km to 5.4 km. Meanwhile, the radius of curvature (Rc) remained stable at 1.4 km, while channel width (W) fluctuated between 0.42 and 0.80 km, narrowing before abandonment.
Between 1635 and 1688 CE, the meander rotated approximately 50° and migrated 1.4 km southwards, with a further 0.3 km migration recorded before flow cessation (Fig. 3). Around 1700 CE, a chute cutoff formed the Isle of Saint-Bertrand (Fig. 3C, Supplementary Fig. 2). Later eighteenth-century CE maps depict flow diverted into the Canal des Launes (thick fuchsia dashed line in Fig. 1), relegating the Bras de Fer to a secondary channel (Supplementary Fig. 3) until its fossilisation (Fig. 3D). By 1754 CE, the Bras de Fer had become reduced to ∼1 m depth and 32 m width at the northern meander (Supplementary Table 1).
The Rc/W ratio fluctuated between 1.8 and 3.3, peaking in 1678 CE, stabilising around 2.2–2.3 (1688–1690 CE), and declining to 1.8 by 1695 CE. It rose sharply after the chute cutoff and avulsion. Migration rates were negligible until 1695 CE, then surged briefly before decreasing again post-cutoff. Apex extension rates were positive in the late seventeenth to early eighteenth centuries CE, turning negative after partial abandonment (Supplementary Table 2).
The bankfull discharge (Qb), along with the most probable annual flood (Q 1.58) also increased during the late seventeenth and early eighteenth centuries CE, reaching Qb ∼16,147 m3/s (1699 CE) and Q 1.58 ∼44,995 m3/s (by 1734 CE), estimated using Dury’s equations. The chute cutoff development led to a decline, with Qb dropping to 11,806 m3/s in 1706 CE (Supplementary Table 2). Given data and method uncertainties, these variations represent indicative trends rather than significant changes.
Detection of overflow forms on the outside bank of the Grande Ponche
The high-resolution DTM of the Grande Ponche meander’s outside bank reveals extensive crevasse splay deposits at Amphise, Pèbre, and Pont de l’Aube (Fig. 4).
Crevasse splay deposits along the outside bank of the Grande Ponche meander (Bras de Fer) on a 50-cm high-resolution digital terrain model (DTM) (IGN, 2022). Deposit extents outlined in black; areas anthropogenically levelled indicated by diagonal dashed lines at Amphise, Pèbre, and Pont de l’Aube.
At Amphise, crevasse splays are broad, with a main channel and few secondary channels. At Pèbre, they are narrow, coalescing, and dendritic (Fig. 4). At the Pont de l’Aube (Fig. 5A), crevasse splays are wider, less elongated, and associated with a complex distributary network. Proximal deposits along the outside bank are coarser, while distal deposits are finer and flatter, marking the transition to the floodplain (Fig. 5A).
(A) A 50-cm high-resolution digital terrain model (DTM) showing the topography of overbank deposits on the outside bank at the Pont de l’Aube, with locations of trenches, electrical resistivity tomography (ERT) profiles, and results of granulometric sampling indicating the coarsest sediments at each site. (B) ERT profiles across lateral overbank deposits of the Grande Ponche meander: one along the levee to crevasse splay transition (A) and three perpendicular to the main crevasse splay channel (B–D). Profile locations shown in A. Each profile has a unique resistivity legend to preserve interpretative accuracy. Standardizing these legends would obscure significant variability;. NGF, Nivellement Général de la France (French national vertical datum). (French national vertical datum)
The DTM (50 cm resolution) further identifies microtopographies within the crevasse splays, such as channels, micro-levees, and channel lobes developing within depressions (Fig. 5A). Vegetation differences further aid discrimination: the normalized difference vegetation index analysis from May 2015 Pleiades imagery highlights strong contrasts in chlorophyll activity, with higher values on levees populated by Tamarix gallica and lower in crevasse splays dominated by Salicornia (Supplementary Fig. 4).
ERT surveys (Fig. 5B) corroborate these findings, suggesting distinct resistivity contrasts: higher values on levees, moderate in channels, and lower on floodplains. Each profile uses a distinct resistivity legend to preserve interpretative accuracy, with values considered relative.
Profiles a-d (Fig. 5B) distinguish two main units: an upper, resistive sandy layer and a lower, conductive silty unit. High resistivity corresponds to sandy, dry levees; intermediate values reflect crevasse splays; and the lowest values mark saturated floodplain deposits. These results are consistent with sedimentological observations from cores and trenches (Figs. 6 and 7).
Cores taken from overbank deposits on the outside bank of the Grande Ponche meander, both in the proximal parts, on the levee (AM_1, PB_1, AU_1), and in the distal parts, on the crevasse splays (AM_2, AM_3, AU_2, AU_3, AU_4), as well as two channel cores (BFII, BF2), taken by Arnaud-Fassetta (Reference Arnaud-Fassetta1998). Overflow deposits reveal three fluvial units consistently identified across proximal and distal areas: basal floodplain deposits overlain by levee or crevasse splay sediments, each subdivided into a bedded, heterogeneous facies (phase 1) and a massive, homogeneous facies (phase 2).
Stratigraphic cross sections showing the chronology and grain-size distribution of levee, crevasse splay, and floodplain deposits at Pont de l’Aube (location in Figure 5A). Radiocarbon (14C) dates from trench samples are marked with red stars, calibrated with Calib Rev 8.1.0 (Reimer et al., Reference Reimer, Austin, Bard, Bayliss, Blackwell, Ramsey and van der Plicht2020) using the IntCal20 curve. Radiocarbon dates agree broadly with archival records. A lagoon-to-fluvial transition (943–1026 CE) predates the documented Bras de Fer period of activity, while a later date (1553–1633 CE) precedes the major depositional phase, possibly reflecting reworking or local sediment pulses. NGF, Nivellement Général de la France (French national vertical datum). (French national vertical datum).
Identification of depositional environments based on sedimentological analysis
Sediment cores (Fig. 6) and trench exposures (Fig. 7) reveal three main fluvial facies: floodplain, levee, and crevasse splay deposits, differentiated through grain-size distributions, geochemical ratios (Zr/Rb, Ca/Si, Si/K), and IRM intensity (Figs. 8 and 9).
Core AU_1 from the base of trench 1.A (location shown in Figure 5A), with log showing photograph, radiography, volumetric granulometry, elemental ratios (Zr/Rb, Ca/Si, Si/K), and isothermal remanent magnetisation (IRM) intensity. The figure illustrates the distinction between fluvial (floodplain) and lagoon deposits. NGF, Nivellement Général de la France (French national vertical datum). (French national vertical datum)
Sediment cores AM_1, AM_2, and AM_3 from overbank deposits at Amphise, with logs showing photographs, radiography, volumetric granulometry, isothermal remanent magnetisation (IRM) magnetic field, and elemental ratios (Zr/Rb, Ca/Si, Si/K). Cores record fluvial environments, with basal floodplain deposits overlain by levee (AM_1) and crevasse splay deposits (AM_2, AM_3). The overbank deposits are subdivided into two depositional phases, distinguishable vertically and laterally relative to the Bras de Fer channel. Additional cores are presented available in Supplementary Material 6. NGF, Nivellement Général de la France (French national vertical datum). (French national vertical datum)
Lagoon environment
Beneath fluvial sequences, a basal lagoonal unit, dated between 899–1042 CE (Amphise) and 943–1026 CE (Pont de l’Aube), comprises fine sands (125–500 µm) and bluish clay-silts (Fig. 10). It is characterised by low Zr/Rb and Ca/Si ratios, weak IRM intensities, and high Si/K, indicative of low-energy, siliceous sedimentation (Fig. 8). Mollusc assemblages (e.g., Cerastoderma edule, Bittium reticulatum) and thin laminae further indicate low-energy lagoonal conditions.
Grain-size distributions (in μm) for fluvial overbank facies of the Grande Ponche and underlying lagoon-marine deposits. Error margins are detailed in Supplementary Material 1.
Fluvial environment
Above the lagoonal deposits, overlying fluvial sediments demonstrate spatial and stratigraphic heterogeneity, reflecting flow dynamics.
Floodplain deposits. Floodplain facies mainly comprise laminated clay-silts with subhorizontal to gently inclined bedding (<15°) (Figs. 6 and 7). They exhibit low Zr/Rb and Si/K ratios, reduced IRM intensities, and slightly elevated Ca/Si ratios compared with other facies (Figs. 8 and 9), attributed to diminished silica-rich input and relative calcium enrichment of detrital origin, given the lack of ostracods. Radiographic images reveal thin graded layers and minimal bioturbation (Figs. 8 and 9). The fine-grained texture and high water content correspond to low resistivity (<1.5 ohm·m) in ERT profiles (Fig. 5B). Fine laminations, colour, oxidation traces, and fauna distinguish floodplain deposits from lagoonal facies (Fig. 7).
Overbank deposits: levee and crevasse splay. Overbank deposits vary with channel proximity. Proximal levee deposits (core AM_1 [Fig. 9]; trenches 1.A and 1.B [Fig. 7]) display inverse grading, current ripples, and oblique bedding enriched in coarse silt (15.6–31.5 μm) (Figs. 5A, 7, and 10), with elevated Zr/Rb and Ca/Si ratios (Fig. 9), reflecting high-energy floods and detrital carbonate input. Si/K ratios increase in sandy intervals and decline in finer layers, indicating hydrodynamic variations (Fig. 9). Distal crevasse splay deposits (cores AM_2 and AM_3) are finer grained and laminated, and show a downstream decline in elemental intensities derived from XRF measurements, consistent with progressive sedimentary fining away from the channel (Figs. 5A, 7, and 9). The absence of ostracods throughout confirms an azoic depositional environment, with carbonate inputs considered detrital.
Both levee and crevasse splay deposits can be subdivided into two depositional phases:
1. Phase 1 consists of coarser, heterogeneous beds with frequent inverse grading and a broad grain-size range (fine silt to fine sand) (Figs. 5A, 7, and 10). This phase is marked by higher Zr/Rb and Ca/Si ratios (Fig. 9).
2. Phase 2 comprises finer, more homogeneous silt-dominated beds (Figs. 7 and 10), with markedly lower Zr/Rb and Ca/Si ratios (Fig. 9). IRM intensities increase slightly relative to phase 1 (Fig. 9), due to the relative enrichment in fine magnetic minerals as coarse quartz and carbonate particles diminish.
Although ERT profiles (Fig. 5B) do not resolve both phases precisely, they distinguish higher-resistivity levees, intermediate crevasse splays, and low-resistivity floodplain units.
Sub-environments of crevasse splay deposits
Crevasse splays are complex internal structures, including shallow distributary channels and small channel–levee systems (Figs. 5A and 7). Several stacked channel fills alternate coarse sand and silt sequences, overlain by flood-silt drapes and separated by subtle erosional contacts (Fig. 7).
ERT profiles occasionally detect near-surface crevasse channels as localised resistivity minima, although deeper lags are rarely discernible due to thinness and lithological similarity to surrounding deposits (Fig. 5B).
The crevasse channels are dominated by fine sands (125–250 μm), with smaller peaks in medium and coarse silts (Figs. 7 and 10). Channel–levee units contain the coarsest material, with more than 40% fine sand in phase 1, while phase 2 becomes more silt dominated (Fig. 10).
Climbing ripples and other current structures, typical of crevasse splay environments (Gulliford et al., Reference Gulliford, Flint and Hodgson2017), were observed in trenches, with gentle dips towards the northeast–southeast (Fig. 7).
Sedimentary structures and geochemical signals collectively denote repeated episodes of large overbank flooding, followed by waning flow phases across the outside bank of the Grande Ponche meander.
Characterisation of depositional processes on the outer bank of the Grande Ponche meander: geochemical and magnetic signatures and transport dynamics
The k-means clustering results, projected onto the principal component analysis (PCA) plane, reveal four distinct groupings that correspond to specific sedimentary facies (Fig. 11). The k-means classification divides samples into clusters based on similarity (Hartigan and Wong, Reference Hartigan and Wong1979), while PCA reduces data dimensions to point out key variations (Jolliffe and Cadima, Reference Jolliffe and Cadima2016). By applying PCA to the k-means results, we reduce high-dimensional geochemical data to two dimensions, enabling clear visualization of the clusters. Lagoonal–marine sands are well isolated in cluster 2, showing clear geochemical distinction from fluvial facies. Floodplain deposits dominate cluster 4, while clusters 1 and 3 contain a mix of levee and crevasse splay samples, reflecting their compositional overlap. Most misclassifications occur among fluvial facies, where certain laminae share similar characteristics.
The k-means classification of sedimentary facies (floodplain, lagoon, levee, and crevasse splay) based on geochemical ratios (Zr/Rb, Ca/Si, Si/K) and isothermal remanent magnetisation (IRM) magnetic field intensity, projected onto the principal component analysis (PCA) PCA plane. This figure presents the classification of U-channel samples collected at 2 mm intervals from overflow deposits in the Grande Ponche area. Symbol shapes indicate facies identified through previous sedimentological analyses. Colours represent four k-means clusters derived from geochemical ratios and IRM intensity: floodplain (cluster 4), lagoon (cluster 2), levee (cluster 1), and crevasse splay (cluster 3). The PCA projection shows that PC1 (x-axis) is driven by Si/K and Zr/Rb ratios, with moderate input from Ca/Si, while PC2 (y-axis) reflects mainly IRM intensity and moderate Ca/Si. The inset correlation circle highlights the geochemical contributions to component separation and facies differentiation.
The Passega CM diagram (Fig. 12) further highlights two main depositional modes at the Grande Ponche outer bank. Floodplain and phase 2 of crevasse splay and levee deposits plot in the T and lower RS fields, indicating low-energy suspension settling. Phase 1 crevasse splay samples extend into the upper RS and QR fields, indicating higher-energy flood events. All facies occur within the RS field, while samples near the QR boundary mark crevasse channel and phase 1 crevasse splay deposition.
Relationship between coarsest and median grain size (C/M) in the outer bank of the Grande Ponche meander. The diagram illustrates sediment transport and depositional processes across different overbank facies, compared with previous studies (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta and Provansal1993, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Arnaud-Fassetta, Reference Arnaud-Fassetta1998, Reference Arnaud-Fassetta2002, Reference Arnaud-Fassetta2013; Vella et al., Reference Vella, Tomatis and Sivan2014). Facies are represented by coloured circles within segments corresponding to specific depositional mechanisms. Three deposition modes are identified: unit 1, overbank-pool facies (suspension, T zone); unit 2, uniform suspension zone (RS); unit 3, graded suspension zone (QR).
Interpretation
An analysis of historical cartography, sedimentary sequences, geophysical surveys, and geochemical proxies demonstrates that the Grande Ponche meander underwent rapid lateral migration and important overbank deposition between 1660 and 1710 CE. Three key observations underpin this interpretation:
1. Hydromorphological instability triggered by climatic variability.
Historical maps highlight an acceleration of meander migration (Fig. 3, Supplementary Table 2), up to 26 m/yr between 1684 and 1700 CE (Fig. 13), coinciding with peaks in flood frequency (Fig. 14). The decline of the Rc/W ratio (Fig. 14) below the stability threshold (Bagnold, Reference Bagnold1960; Hey, Reference Hey1984; Hooke, Reference Hooke1997), together with the documented outside bank accretion (Fig. 4), reflects instability and a transition towards the chute cutoff.
Figure 13.“Vue figurée de la division des acretements du Rhône entre la comté d’Arles et le tenement de Gouine” (Source: AC Arles - 1Fi282, Arles municipal archives: https://arles.fr/decouvrir/les-archives-communales/consultation-des-archives-communales/). (A) Localised map around La Vignolle showing two former Rhône shorelines (1684 and 1700 CE), with an estimated inside bank gain of 215 “compas” (∼418 m). (B) Image of the area likely corresponding to the tract indicated on the Vignolle map, identified through landscape similarities and distances between former shorelines mapped in 1688 and 1706 (points B and H in A).
Figure 14.Hydromorphological evolution and flood frequency (Pichard, Reference Pichard1995) of the Grande Ponche meander (1630–1739 CE) in relation to Atlantic Multidecadal Oscillation (AMO) phases (Gray et al., Reference Gray, Graumlich, Betancourt and Pederson2004). Flood frequency peaks during AMO+ phases, suggesting a climatic influence. Morphometric parameters (Rc/W declining; amplitude, sinuosity, and wavelength increasing) indicate meander instability and progression towards abandonment. Estimated bankfull discharges (∼16,000 m3/s) exceed modern Rhône flood values, reflecting extreme hydrological conditions.
Application of Dury’s formula (Fig. 14) provides bankfull discharge estimates above 10,000 m3/s for the Grande Ponche meander, consistent with the Rhône’s largest historical floods (12,500 m3/s in 1856; 11,500 m3/s in 2003; Levraut and Roy, Reference Levraut and Roy2007) and far exceeding the Rhône’s present-day mean annual discharge (∼1680 m3/s; Hydro.eaufrance, n.d.). Deforestation and land-use changes (Bravard, Reference Bravard1989, Reference Bravard2010) likely enhanced sediment flux.
This frequent flood period aligns with positive phases of the Atlantic Multidecadal Oscillation (AMO+) (Gray et al., Reference Gray, Graumlich, Betancourt and Pederson2004; Mann et al., Reference Mann, Zhang, Rutherford, Bradley, Hughes, Shindell and Ammann2009; Knudsen et al., Reference Knudsen, Jacobsen, Seidenkrantz and Olsen2014), characterised by increased precipitation extremes and altered wind patterns (Curtis, Reference Curtis2008; O’Reilly et al., Reference O’Reilly, Woollings and Zanna2017). The synchronicity between AMO+ phases and heightened Rhône fluvial activity (Fig. 14) supports the influence of decadal climatic oscillations, potentially amplified by the Maunder Minimum and enhanced volcanism (Miller et al., Reference Miller, Geirsdóttir, Zhong, Larsen, Otto‐Bliesner, Holland and Bailey2012; Brönnimann et al., Reference Brönnimann, Franke, Nussbaumer, Zumbühl, Steiner, Trachsel and Hegerl2019) around 1700–1710 CE.
2. Extensive crevasse splay deposits indicative of multiple bank breaching.
Crevasse splay deposits detected along the outside bank reflect repeated natural levee breaches during large floods (Fig. 4). As detailed in “Identification of Depositional Environments Based on Sedimentological Analysis,” the overbank deposits exhibit a dual-phase organisation: phase 1, comprising coarse, heterogeneous beds deposited under high-energy conditions during flood surges; and phase 2, consisting of finer, more homogeneous silts laid down during declining flow energy. Elevated Zr/Rb and Ca/Si ratios in phase 1 confirm intense detrital input during peak flooding. Their decline in phase 2, along with a slight increase in IRM values, points to finer sedimentation and a relative enrichment in magnetic minerals as coarse inputs diminish.
These facies may represent either distinct floods or the front- and back-loading phases of singular flood events, as suggested by Filgueira-Rivera et al. (Reference Filgueira-Rivera, Smith and Slingerland2007). Fine-grained phase 2 deposits may also reflect channel abandonment processes or distal overbank sedimentation (Gulliford et al., Reference Gulliford, Flint and Hodgson2017).
The geometry of the crevasse splays (lobe-shaped, <250 m wide, and <200 m long) and their sedimentary organisation correspond to distributive, non-channelised flow typical of low-gradient floodplains (Coleman, Reference Coleman1969; Gulliford et al., Reference Gulliford, Flint and Hodgson2017; Millard et al., Reference Millard, Hajek and Edmonds2017). The modest extent and silty composition of Grande Ponche crevasse splays reflect the fine sediment input and low-gradient floodplain.
Rapid lateral channel migration, high discharges, and bar instability likely increased overbank flooding and crevasse splay formation (Coleman, Reference Coleman1969). The deposits at Grande Ponche are interpreted as products of these processes. While individual crevasse splays elsewhere may account for 60–70% of overbank stratigraphy (Burns et al., Reference Burns, Mountney, Hodgson and Colombera2017) and form complexes up to 221 km2 (Rahman et al., Reference Rahman, Howell and MacDonald2022), those at Grande Ponche form a stacked succession over ∼5 km2 along a migrating meander.
3. Magnetic and elemental proxies indicating abrupt pulses of detrital sediment input.
Magnetic (IRM) and elemental (Zr/Rb, Ca/Si, Si/K) proxies record discrete pulses of detrital influx linked to flood events. Low IRM values align with coarse-grained overbank deposits, while sustained high IRM intensities correspond to finer-grained, floodplain sedimentation (Fig. 9). Strong magnetic signals occur where coarse diamagnetic quartz is scarce and ferromagnetic minerals, transported within clay fractions, predominate. Conversely, low IRM values and low Si/K ratios indicate dilution by quartz-rich material.
Elevated Zr/Rb and Ca/Si ratios during phase 1 reflect coarse sediments from the Massif Central and Rhône Basin limestone catchments (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta, Provansal, Garnier and Mouchel1999; Stanley and Jorstad, Reference Stanley and Jorstad2002). Their decline in phase 2, alongside rising IRM intensities, marks a transition to finer sedimentation under lower-energy conditions. Zr/Rb ratios peak in AM_1 and AM_2, near the levee breach, reflecting a coarse Zr-bearing sediment input, while AM_3, more distal, records finer-grained material (Fig. 9).
Sedimentary characteristics, such as the absence of bioturbation, the presence of graded bedding, and the geochemical signals (elevated Zr/Rb and Ca/Si in phase 1; high IRM in phase 2), indicate rapid deposition.
The k-means clustering based on elemental ratios and magnetic signatures (Fig. 11) confirms four distinctive facies (floodplain, lagoon, levee, and crevasse splay). Lagoonal facies form a well-defined and distinct cluster, whereas fluvial facies show more overlap. Confusion is frequent between levee and crevasse splay facies (clusters 1 and 3). These overbank deposits are polyphased and share similar compositions. They can include fine silty-clayey laminae deposited during waning flood stages, leading to geochemical signatures that partly resemble those of floodplain deposits. Topographic context helps differentiate distal crevasse splays from the higher-elevation levees. Moreover, low IRM values typically align with coarse-grained overbank layers and channel-proximal settings, while consistently elevated IRM intensities characterise fine-grained, magnetically enriched floodplain sediments.
Our findings demonstrate that short-term climatic variability exerted a control on the fluvial dynamics of the Rhône delta during the late LIA, driving rapid meander migration and overbank accretion. The following discussion situates the Grande Ponche case within a broader geomorphological and palaeoclimatic framework, considering the implications for deltaic systems under hydroclimatic forcing.
Discussion
Climatic forcing and rapid fluvial adjustment
The late LIA evolution of the Grande Ponche meander demonstrates that 1650–1710 CE climatic variability, likely associated with an AMO+ phase, triggered extreme floods and rapid geomorphic adjustments in the Rhône delta. Historical cartography and stratigraphy reveal significant meander migration and widespread crevasse-splay deposition over more than 5 km2, reflecting a high sensitivity of deltaic rivers to decadal-scale hydroclimatic forcing.
While a study from the Mississippi suggests stronger flood events during AMO− phases (Munoz et al., Reference Munoz, Giosan, Therrell, Remo, Shen, Sullivan and Wiman2018), our results highlight a contrasting regional response, where AMO+ appears to coincide with intensified flooding. This emphasises the importance of local climatic–hydrologic coupling in determining fluvial dynamics.
A strengthened summer North Atlantic Oscillation may have further contributed to increased runoff in the Rhône catchment (Schulte et al., Reference Schulte, Peña, Carvalho, Schmidt, Julià, Llorca and Veit2015).
Sedimentary records of intense flood pulses
The underlying lagoonal sediments, identified at −0.1 m NGF (Nivellement Général de la France, French national vertical datum) at Amphise and between −0.25 and −0.65 m NGF at Pont de l’Aube, are consistent with observations at Saint-Bertrand (Arnaud-Fassetta, Reference Arnaud-Fassetta1998) and indicate that the present fluvial landscape evolved through the infilling of former coastal environments during delta progradation. The presence of Cerastoderma edule and Bittium reticulatum further attests to low-energy lagoonal conditions, as similarly recorded near the Ulmet palaeochannel (Vella et al., Reference Vella, Landuré, Long, Dussouillez, Fleury, Tomatis, Sivan, Sanchez and Jézégou2016).
Units on the outside bank (Fig. 15) partially align with channel fill inside bank sequences described by Arnaud-Fassetta (Reference Arnaud-Fassetta1998). Early deposits (Unit 1) likely record distal events before major accretion (1695–1711 CE). Subsequent phases match Channel Units I–III, suggesting late-stage floods enhanced overbank buildup. Post-1711 avulsion reactivated parts of the channel (Units IV–V) but had limited impact on distal deposits.
Conceptual diagram of stratigraphy and sediment textures of overbank and channel deposits in the Grande Ponche meander. Overbank units partially correspond to sedimentary core data from palaeochannels documented by Arnaud-Fassetta (Reference Arnaud-Fassetta1998); The floodplain (overbank unit 1) likely represents distal deposits from older channel systems predating the 1695 CE meander expansion. Cores BF2 and BFII reveal three phases of channel activity: (1) coarse sands from high-energy bars; (2) finer deposits from reduced flow; and (3) moderately coarse, thick sands. These phases correspond to two overbank units: Unit 2 (proximal sandy levee and distal sandy silt crevasse splay) and Unit 3 (proximal sandy silt levee and distal silty crevasse splay).
The CM diagram (Fig. 12) allows comparison with sediment transport dynamics across the Rhône delta. Unlike the spatial sorting typical of fluvial systems (Passega, Reference Passega1977), no clear grain-size gradient is observed at the Grande Ponche outer bank. This likely reflects local depositional controls—short transport distances, repeated sediment reworking—and possible sampling limitations.
Similar patterns have been reported elsewhere in the delta, where overbank and channel deposits often plot within the same CM fields, particularly the RS domain, despite granulometric differences (Arnaud-Fassetta, Reference Arnaud-Fassetta2002; Vella et al., Reference Vella, Tomatis and Sivan2014). This overlap suggests a broad range of flow competence and supports the interpretation of hydrodynamic continuity between channel and overbank processes.
The Grande Ponche sediments are broadly comparable to floodplain deposits dated between 2350 and 1550 BP at Saint-Ferréol, Ulmet, and Peccais (Arnaud-Fassetta, Reference Arnaud-Fassetta2002). At Ulmet, levees are finer, while channel fills range from similar to coarser than overbank deposits at Bras de Fer (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta and Provansal1993; Vella et al., Reference Vella, Tomatis and Sivan2014).
Post-1711 CE crevasse splay and levee deposits in the Grand and Petit Rhône are generally finer (Arnaud-Fassetta and Provansal, Reference Arnaud-Fassetta and Provansal1993), whereas the 2003 flood deposits are as coarse—or coarser (Arnaud-Fassetta, Reference Arnaud-Fassetta2013)—than those at Grande Ponche (Fig. 12), indicating high-energy deposition likely linked to artificial levee breaches and concentrated overbank flows.
All Grande Ponche samples were pretreated to remove organic matter and carbonates, ensuring accurate characterisation of the detrital fraction. Earlier studies may not have followed the same protocol, potentially inflating apparent grain sizes. Methodological differences may thus partly explain intersite variability.
The Grande Ponche crevasse splays, despite numerous similarities, exhibit distinct characteristics compared with other examples. Their small size and simple architecture contrast with the complex, amalgamated crevasse splays described by Burns et al. (Reference Burns, Mountney, Hodgson and Colombera2017) in Cretaceous systems and by Gulliford et al. (Reference Gulliford, Flint and Hodgson2017) in fine-grained floodplain successions, likely reflecting differences in channel scale, sediment supply, and floodplain accommodation.
Unlike large crevasse splays in broad floodplains (Coleman, Reference Coleman1969; Aslan et al., Reference Aslan, Autin and Blum2005), the Grande Ponche deposits reflect a confined environment limiting sediment dispersal. The dominance of silt and the scarcity of fine sand, in contrast to coarser examples (O’Brien and Wells, Reference O’Brien and Wells1986; Millard et al., Reference Millard, Hajek and Edmonds2017), highlight the role of local hydrosedimentary conditions.
The persistence and recurrence of crevasse splays at Grande Ponche, unusual in European river systems, point to a strong interplay between LIA climatic forcing and Rhône-specific fluvial dynamics.
Implications for deltaic environments
Holocene sea-level changes in the western Mediterranean were complex, shaped by climatic, isostatic, and tectonic factors (Di Donato et al., Reference Di Donato, Negredo, Sabadini and Vermeersen1999; Calafat et al., Reference Calafat, Frederikse and Horsburgh2022; Marriner et al., Reference Marriner, Kaniewski, Morhange, Vacchi, Kallel, Rossignol and Cazenave2023). During the LIA, relative sea-level rise slowed considerably (Vacchi et al., Reference Vacchi, Ghilardi, Melis, Spada, Giaime, Marriner and Lorscheid2018, Reference Vacchi, Joyse, Kopp, Marriner, Kaniewski and Rovere2021), allowing riverine sediment inputs to dominate delta dynamics.
These trends were modulated by the AMO and Atlantic Meridional Overturning Circulation, influencing North Atlantic heat transport and Mediterranean levels (Marriner et al., Reference Marriner, Kaniewski, Morhange, Vacchi, Kallel, Rossignol and Cazenave2023). Stabilised sea levels during the LIA varied spatially due to atmospheric pressure, sterodynamic effects, and glacial ice loss (Calafat et al., Reference Calafat, Frederikse and Horsburgh2022). Prevailing wet conditions contrasted with earlier dry phases (Fletcher and Zielhofer, Reference Fletcher and Zielhofer2013). Glacio-hydro-isostatic adjustments, enhanced by distance from former ice sheets, also modulated sea levels, particularly in tectonically stable areas like the southern Peloponnese (Lambeck, Reference Lambeck1995).
Although marine influence during the LIA was limited, it remains crucial for interpreting delta evolution. The Bras de Fer prograded by ∼1.5 km between 1635 and 1678 CE (authors of this study, unpublished data), despite the modest sea-level rise (0.45 ± 0.7 to 0.6 ± 0.6 mm/yr; Vacchi et al., Reference Vacchi, Joyse, Kopp, Marriner, Kaniewski and Rovere2021). Simultaneously, the Grande Ponche meander underwent significant vertical aggradation, reflecting enhanced fluvial input and overbank sedimentation.
Recent Mediterranean sea-level rise has accelerated, reaching 3.6 ± 0.4 mm/yr (2000–2018), driven by warming and ice melt (Calafat et al., Reference Calafat, Frederikse and Horsburgh2022). Projections estimate a rise to ∼4.2 mm/yr by 2040–2050 (Galassi and Spada, Reference Galassi and Spada2014), and potentially ∼6–10 mm/yr by 2100 under high-emission scenarios (Lionello and Scarascia, Reference Lionello and Scarascia2018). These rates exceed those of the LIA, increasing risks of shoreline retreat and salinisation. Sustained sediment delivery remains essential for delta resilience (Fu et al., Reference Fu, Xiong, Zong and Huang2020; Curtis et al., Reference Curtis, Flint, Stern, Lewis and Klein2021), partially mitigating sea-level rise impacts.
The Grande Ponche case illustrates fluvial sensitivity to short-term climatic forcing, although extrapolation to other deltas must be cautious. Observed dynamics—meander migration, overbank accretion, crevasse splay formation—reflect local sediment supply, floodplain morphology, and hydroclimate. These processes are not directly applicable to wave- or tide-dominated systems. Broader generalisation requires comparative studies across diverse deltas. This study thus enhances understanding of climate–fluvial interactions in deltaic plains and highlights the need for further research on variability and thresholds across settings.
Conclusion
This study reconstructs the rapid evolution of the Grande Ponche meander during the LIA, illustrating how short-term climatic oscillations influence fluvial dynamics in deltaic settings. Using a multiproxy approach combining historical cartography, topography, sedimentology, geochemistry, and geophysics, we show that episodes of more frequent flooding, associated with positive AMO phases, triggered rapid meander migration and extensive overbank deposition between 1650 and 1710 CE.
Our findings reveal that climatic oscillations promoted hydromorphological instability, leading to significant sediment redistribution through crevasse splay deposition. Sedimentary, geochemical, and magnetic proxies document abrupt pulses of detrital input by large floods, recording the sensitivity of the Rhône delta to decadal-scale hydroclimatic forcing despite relatively stable sea levels.
The Grande Ponche case study underscores the capacity of deltaic environments to respond to short-term climatic fluctuations. It also illustrates the value of multiproxy approaches for reconstructing past fluvial processes at fine (decadal) temporal and spatial resolution. These findings contribute to a broader understanding of climate–fluvial interactions and offer key insights for predicting the future resilience of deltaic systems in the context of accelerated sea-level rise and anthropogenic pressures.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/qua.2025.10031
Acknowledgments
We thank the Réserve Naturelle Nationale de Camargue and the Parc Naturel Régional de Camargue, especially Mr. Sylvain Ceyte from the PNRC Natural Areas Management Unit, for logistical support and site access. This study was funded by OHM Vallée du Rhône and the PCR–Ministry of Cultural Affairs, SRA Aix-en-Provence. We are grateful to Charlotte Yonnet and the Conservatoire du Littoral for field access, and to Pauline Vitry for her work on sediment sample calibration. Thanks also to Sophie Viseur for supervising the granulometric analyses, to be published in more detail in a future study. M.S.N. Carpenter post-edited the English style and grammar.
Open access
